Teraminutes for the meeting of Friday April 30th 2004

Scamp, astrometrix (Mireille, Yannick, Henry).

 Scamp doesn't appear to be fully functioning yet. For example, based on the tests Mireille has carried out, adding all images together for a given field produces a very bad astrometric solution: the internal errors are sometimes about 20 times larger than we need to coadd the images.

 However, taking observations only from the same night, the astrometric solution is much better. It is not clear *how* to select images to get a good astrometric solution (moreover, the suggestion from Emmanuel, to separate observations based on QRUNID doesn't seem to work).

 Finally, tests carried out by Henry on the COSMOS i* data indicate Scamp cannot produce yet an acceptable astrometric solution for images from the same night taken at approximately the same airmass can. The only case where Scamp seems to function well for images taken very close together in time.

 Mireille showed some rebinned images of the pre-survey fields (14 deg2), swarped used an astrometric solution computed by astrometrix. Seems to work very well, once the images with a large amount of extinction are removed.

 An important point raised by Yannick is that it's not clear that Scamp will be ready for the Users' meeting in a little over a week. In which case we should concentrate on running 'astrometrix/photometrix' on the d3, w3, pre-w3, d1, W1, pre-w1, fields. (This will all reduce pressure on Emmanuel/Laurent).

To summarise:

 This processing will be carried out on mix2/mix1.
 Henry and Chiara for D3/D1 (mix2) and
 Mireille and Yannick for W3/PreW3 and W1/PreW1 (Gilles continues work on the W2 data set).
 Henry will first check Photometrix is running/working on mix1.
 Fred will install astrometrix/photometrix on mix1.

 Astrometrix will be used until Scamp is fully validated. Adopting this strategy will reduce the pressure on Emmanuel who can concentrate debugging Scamp without trying to meet an unrealistic deadline.

 Fred will look at the astrometrix source code to see if it can be modified to directly read the LDACs generated by QualityFITS: the big drawback with astrometrix is that it works on split MEF files, and runs SExtractor, which is very very time-consuming. We need to investigate if we can eliminate this step.

 at the same time, Yannick/Henry will continue to test Scamp.

Photometric calibration in Scamp: Mireille thinks at the moment that there might be a slight underestimation in the photometric re-scaling. Although for the pre-wide survey images, the internal photometric residuals are small.

QualityFITS (Fred, Laurent): A few images (5) have problems with background generation. This comes from a bug in convert (from ImageMagick), which fails when it tries to read the TIFF files generated by STIFF. This is probably due to a bug in the TIFF library. There are 5-6 images which have this problem; this is only a problem for web-page generation / visualisation.

 Furthermore, 150 images have weight-maps in which the CRVALs which are missing, as above, this is only a problem for visualisation purposes. To fix this we need just to copy the correct CRVALs back into the weight-maps and then reswarp the images.

Speeding up QualityFits report page (Laurent:)

 We will wait until all images have been processed by QualityFits before investigating this detail.

SkyProbe data status(Jean-Christophe:)

 Should have all the data early next week. The data we have received has some problems with the time order of the data-points. CFHT/CADC have been contacted.

QF2 and high level quality assessments (Laurent, Fred)

 Fred will carry out statistics (for example, seeing) on 7'x7' selected areas to approximately cover each of the CCDs on input. Will produce catalogues for Laurent, for the higher-level quality control.

SPICA (Laurent):

 SPICA has problems if there is only a single image to be processed (the seeing criterion is very restrictive and many images are rejected). Will send all QualityFITS'ed images to the CADC next week. We will try to use bftp to do this (Fred is not convinced that the bftp protocol give much better performance than the normal ftp protocol).

Skywatcher (JC):

 Skywatcher now shows the galactic plane in green, path of the ecliptic in orange. Different fields are coloured differently. Still waiting for feedback in order to compile le HTM library. Next step is image splitting according to filter and RUNid or survey types in several sky maps.

Dbclient2 (Gerard, Fred)

Testix re-installation not finished yet, and so no dbclient2 tests have been carried out.

Panier (Gerard):

 Selections using the Qfits parameters are implemented. New version should be delivered prior leaving to Groningen.

AVOcat (Gerard): Hope to send to version to CDS today/monday. Will be presented at the Groningen meeting, next Wednesday.

Panorapix (Fred):

 It's not yet possible to compile Panorapix in 64bit mode on the Opterons...

Software installation on the Opterons (Henry, Fred):

 All important 32 bit astronomical applications we use (iraf, skycat, gaia) now work on the Opterons thanks to Fred installing the necessary 32 bit compatibility libraries. In future Henry will tell Fred when he installs software on the mix2.

Lists:

 Fred suggests that we provide a list of tags for each subject for mails sent to the TERALIST so that it's easier to find mails in the archive. Gerard proposed a list of tags ..e, [truc][TERALIST].

Hardware (Fred):

 /raid1 on mix2: some more tests with cables required...

 Need to ask for three 'devis' for an Opteron bi-pro to add to pix(1-9), essentially around 9000 Eur.

 Gilles: perhaps we can salvage 1Tb of disk from the alphas to use as storage for incoming/outgoing data on clix. This would be a 'low speed' disks (approximately 40mb/second), suitable for near-line storage. This TByte will be added to the CDAC/Terapix incoming/outgoing data storage before the end of the year.

Next Tera-meeting: Friday 7th May 2004